Brookhaven National Laboratory´s Spar Buoy Oceanographic Telemetry System (SBOTS): Software Engineering in an Oceanic Information System

Since May 1977 over a quarter of a million in situ measurements of temperature, conductivity, and current have been collected off the south shore of Long Island by a real-time data collection system. The size of this data set by early 1978 and plans for its continued expansion were deemed sufficient justification for undertaking an intensive information system design effort beginning. The goals of the SBOTS information system are (1) to significantly decrease the lag time between spar deployment and the availability of reliable information for distribution and analysis; and (2) to provide a set of programs flexible enough to incorporate both changes in data verification strategies and sensor package hardware enhancements with minimal reprogramming. These goals have been met by using the techniques of Software Engineering, including a top-down design, a hierarchical approach to program definition, structured programming, and IPO diagram documentation. The top-down approach to the information system design is illustrated by a series of block diagrams describing both high-level systems definition and detail programming task definition. Programming paralleled the top-down design. The use of the University of Minnesota FORTRAN Compiler to obtain structured programming and library functions to obtain modularity are discussed. Several levels of IPO diagrams are included as examples of the SBOTS system documentation. A unique feature of the SBOTS information system is the use of statically structured data files. Several applications which take advantage of this feature are described, including the use of digital filters and Fourier methods for time series analysis. Procedures for pre- and post-deployment calibration of each remote sensor have been established and now form an integral part of the SBOTS information system. Internal calibration techniques to compensate for electronic drift in the sensing and telemetry system have been refined and both are described. The softwa- - re measures of development cost, complexity, maintainability, and coding effort are compared between the programming prior to mid-1978 and since.